How to Manage an Underperformer Without Losing Them

Two managers, same situation. Both had a direct report whose work had quietly slipped over a few months. Missed details, slower output, a couple of dropped balls that landed on other people.

Marcus, the first manager, did what felt natural. He waited. He didn't want to make it a thing. He picked up the slack himself, told himself it was a rough patch, and said nothing for two quarters. Then performance review season hit, the gap was now undeniable, and he went straight to a formal warning. His report was blindsided. She'd had no idea anything was wrong, because nobody had told her. Within a month she'd left, angry, and two of her teammates started updating their resumes because they'd watched how it went.

Devon, the second manager, did the harder thing first. Three weeks into noticing the pattern, he sat his report down and said, plainly and kindly, "Your work has slipped lately and I want to understand why before it becomes a bigger problem." That one conversation surfaced something Devon never would have guessed. His report was caring for a sick parent and drowning. They built a plan together. Six weeks later the work was back. A year later that person was one of his strongest performers.

Same problem. Opposite outcomes. The difference wasn't talent or toughness. It was that one manager treated underperformance as a verdict to deliver, and the other treated it as a problem to solve. This post is about being the second manager.

Wait too long, then move too fast

The most common pattern in managing underperformance is also the most damaging: silence, then a sudden escalation.

It happens because the early conversation is uncomfortable and the evidence still feels arguable. So you wait. You tell yourself you need one more data point. Meanwhile the person has no idea they're falling short, the rest of the team is absorbing the cost, and your own resentment is building quietly in the background. By the time you finally act, you're frustrated enough to skip straight to consequences, and the person experiences the whole thing as an ambush.

Kim Scott, in Radical Candor, frames the job as caring personally and challenging directly, at the same time. Waiting and then escalating fails both halves. The silence feels like kindness, but it's really avoidance, and the person is the one who pays for it. The eventual warning feels like directness, except it lands months after it could have actually helped.

Common mistake: Confusing "I haven't said anything harsh" with "I've been kind." Letting someone fail quietly for months is one of the unkindest things a manager can do. You're denying them the one thing that could fix it: the truth, early enough to act on it.

Diagnose before you decide

Underperformance is a symptom, not a diagnosis. Before you decide what to do, figure out what you're actually looking at. In my experience it's almost always one of these:

1. A clarity problem. They don't actually know what good looks like, or expectations shifted and nobody told them. This is more common than managers want to admit, and it's the easiest to fix.
2. A skill problem. They understand the bar and can't yet clear it. This is coachable, and often just a matter of time and reps.
3. A capacity problem. Something outside work, or too much inside it, is eating the energy the job needs. Devon's report was here.
4. A motivation problem. They've checked out. The role, the team, or the direction stopped mattering to them. This is the hardest, and the one most worth naming directly.
5. A fit problem. The person is capable but genuinely in the wrong seat. Sometimes the kindest outcome really is a different role, or a different company.

You cannot manage these the same way. Coaching someone through a motivation problem as if it were a skill gap wastes everyone's time. The diagnosis determines the plan, so do the diagnosis first.

The way you diagnose is not by guessing. It's by asking, and then being quiet long enough to hear the answer.

Have the conversation early, and make it a conversation

The early conversation is the whole game. Get it right and most underperformance never reaches a formal process at all.

Lead with the specific, not the general. "You've seemed off lately" gives someone nothing to work with and puts them on the defensive. "The last three reports had errors I had to catch, and that's not where your work usually is" gives them something concrete and signals that you remember they're capable of better.

Then ask, and actually mean it: "What's going on?" This is the line Marcus never got to. The answer is where you find out which of the five problems you're dealing with. Resist the urge to fill the silence or to solve it on the spot. You're gathering the diagnosis, not closing the case.

"The conversation you're dreading is the one that, had you held it a month ago, you wouldn't be dreading now."

Try this week: If there's someone whose work has slipped and you've been sitting on it, book 30 minutes and open with one specific, recent example. Not a list. One. Then ask what's going on and let them talk first. You're not delivering a verdict. You're starting a problem-solving conversation while it's still small enough to solve.

Build a plan, not a verdict

Once you understand the cause, you and the person build the path back together. The word "together" matters. A plan you hand down is a sentence. A plan you build with them is a commitment.

A real plan has three things. It names what specifically needs to change, in observable terms, so both of you will recognize success when you see it. It names what support they'll get from you, because expecting different results with zero change in support is just blame with extra steps. And it names a timeframe, so the situation can't drift back into the open-ended fog it came from.

Write it down. Not because you're building a paper trail for HR, but because memory is generous to both parties and a shared document keeps everyone honest. Then you check in on it. Often. A plan you set and never revisit is a plan you've quietly abandoned, and the person can feel that.

"People can recover from almost any honest piece of feedback. What they can't recover from is finding out you'd given up on them weeks before you said a word."

Know where the line is

Not every story ends like Devon's, and pretending otherwise does no one any favors.

Some people, given a fair runway and the support to use it, still don't get there. When you've genuinely done your part and the gap remains, the right move is to part ways, cleanly and respectfully. That's not a failure of the process. That is the process working. The goal of managing an underperformer well was never to save every single person. It was to make sure that whatever happens, it happens with honesty, with a real chance to recover, and without the person being surprised at the end.

Gallup's research keeps finding the same thing about why people leave: it's rarely the work itself, and very often the manager. A botched performance conversation doesn't just lose you the one person. It tells everyone watching exactly how they'll be treated on their worst month. Handle it well and the opposite is true. The team learns that struggling here doesn't mean you're disposable.

I built Cadence partly because this is the kind of thing that falls apart in the gaps. The early concern you meant to raise and didn't. The plan you agreed to and never revisited. It keeps the goals, the 1:1 notes, and the follow-ups in one place, so the hard conversation actually turns into follow-through. But the tool isn't what saves the person. The habit of not letting it drift is.

Be Devon. Say the true thing early, while it's still small enough to fix.

If you want hard conversations to turn into real recoveries instead of paperwork, you can learn more about Cadence at cadencehq.co.

Turn the conversation into a plan.

Cadence gives you a place to track the goals, the check-ins, and the follow-through that turn a hard conversation into a real recovery. 14-day free trial.